氧化铝生产过程苛性比值与溶出率智能集成预测模型研究
详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文
摘要
苛性比值与溶出率是氧化铝高压溶出过程中两个重要的经济技术指标。它们不仅决定了氧化铝溶出的效果及碱耗,而且对氧化铝的后续生产具有极大的影响。然而,目前苛性比值与溶出率的检测严重滞后、波动范围大且难以及时调整,从而导致整个生产流程的实时控制陷入被动。为此,研究如何运用智能集成建模方法建立苛性比值与溶出率的预测模型从而实现苛性比值与溶出率的在线检测,对实现氧化铝生产过程的稳产高产、提高企业竞争力都具有重要意义。
     本文以中国铝业公司河南分公司氧化铝生产过程为背景,着重研究了苛性比值与溶出率的智能集成预测模型的建立和应用。首先,在分析氧化铝高压溶出过程机理的基础上,确定了影响苛性比值与溶出率的主要因素,提出了苛性比值与溶出率的机理模型;然后,提出了基于主元分析的多神经网络模型,用来补偿机理模型的偏差,从而建立机理模型与神经网络的集成模型;接着又提出了具有自校正样本库的基于聚类分析的匹配模型,用来预测苛性比值与溶出率;最后在深入分析机理模型与神经网络集成模型、匹配模型两者的优点与不足的基础上,提出了采用基于专家知识和统计学知识的智能协调器对两者的输出进行协调,从而建立苛性比值与溶出率的智能集成预测模型。
     智能集成预测模型的现场运行结果表明该模型具有较高精度,能很好地实现苛性比值与溶出率的在线预测。
Ratio of Soda to Aluminate (RSA) and Leaching Rate(LR) are two very important economical and technical indices in the process of High-Pressure Digestion (HPD) of alumina. Not only they affect output of alumina and alkali consumption, but also they make great influence on successive production of alumina. However, at present it is difficult to measure RSA and LR timely, which makes it hard to achieve real-time control of the whole process of production. Therefore, How to implement on-line measurement of RSA and LR by establishing prediction model using intelligent integrated modeling, is significant for realizing steady and high production of alumina and enhancing the competitive power of enterprises.
    On the background of production process of alumina in Henan Brach China Aluminium, this paper mainly does research in the establishment and application of intelligent integrated prediction model of RSA and LR. Firstly, the key factors that influence RSA and LR are acquired by analyzing the mechanism of process of HPD, then mechanism model of RSA and LR is established. Secondly, in order to compensate errors of mechanism model, multiple neural networks based on principle component analysis is put forward, thereby integrated model of mechanism model and neural network is proposed. Thirdly, matching model with self-tuning sample library based on cluster analysis is put forward. Finally, on the base of analyzing merits and flaws of integrated model of mechanism model and neural network and matching model, intelligent coordinator based on expert knowledge and statistics is designed to harmonize outputs of these two models, therefore intelligent integrated prediction model of RSA and LR is established.
    An real-world application in Henan Brach China Aluminium shows that intelligent integrated model is effective and it can predict RSA and LR on line very precisely.
引文
[1] 杨重愚.氧化铝生产工艺学.北京:冶金工业出版社,1993.150~170
    [2] 《联合法生产氧化铝》编写组.联合法生产氧化铝:原料制备.北京:冶金工业出版社,1977.25~35
    [3] 王雅琳.智能集成建模理论及其在有色冶炼过程优化控制中的应用研究:[博士学位论文].长沙:中南大学,2001
    [4] 陈晓方,桂卫华,蔡自兴,等.过程控制中的智能集成建模方法.系统仿真学报,2001,13(增刊):8~11
    [5] 王亦文,桂卫华,王雅琳.基于最优组合算法的烧结终点集成预测模型.中国有色金属学报,2002,12(1):191~195
    [6] Wang Yalin, Gui Weihua, et al. A multi-objective optimization for the mixture ratio of Pb-Zn sintering process based on an intelligent integrated model. AMSMA'2000, GuangZhou, 2000
    [7] Chen Xiaofang, Gui Weihua, Wang Yalin. A prediction model of sulfur based on intelligent integrated strategy. Proceedings of IFAC 15th Congress, Barcelona, 2002
    [8] Wan W H, Shi W X, Yuan Y F. Fuzzy modeling of nickel flash smelting process [J]. Metallurgical industry automation, 2000, 24(2):9~12
    [9] Xu G L, Li S P. Study on neural network forecast model of the secondary cooling of continuous casting [J]. Basic automation, 1999, 6(1): 5~7
    [10] Kodkinen J, Yliniemi L, Leivisk K. Fuzzy modeling of a rotary dryer [A]. In: Preprints of the IFAC Workshop [C]. Finland: Elsevier Science, 2000.166~171
    [11] Wu M, Nakano M, She J H. A model-based expert control strategy using neural networks for the coal blending process in an iron and steel plant [J]. Expert Systems with Applications, 1999, 16(6): 271~281
    [12] 陈元青,陈琦,王树青.多元统计分析方法在链霉素发酵中的应用.生物工程学报,1999,15(3):368~372
    [13] 陈耀,王文海,孙优贤,基于动态主元分析的统计过程监视.化工学报,2000,51(5):666~670
    [14] 赵立杰,王纲,李元.非线性主元分析故障检测和诊断方法及应用.信息与控制,2001,30(4):359~364
    [15] 张铃,张钹.人工神经网络理论及应用.浙江:浙江科学技术出版社,1997.30~55
    
    
    [16]姚宏伟,梅晓榕,庄显义.模糊聚类分析在模糊神经网络结构优化中的应用.高技术通讯,2000,10(4):63~66
    [17]单世谱.联合法生产氧化铝:高压溶出.北京:冶金工业出版社,1974.15~20
    [18]张杰,阳宪惠.多变量统计过程控制.北京:化学工业出版社,2000.5~10
    [19]任若恩,王惠文.多元统计数据分析——理论、方法、实例.北京:国防工业出版,1997.20~30
    [20]方开泰.《实用多元统计方法》.上海:华东师范大学出版社,1986.30~45
    [21]杨斌,田永青,朱仲英.智能建模方法中的数据预处理.信息与控制,2002,32(4):380~384
    [22]王科俊,王克成.神经网络建模、预报与控制.哈尔滨:哈尔滨工程大学出版社,1996.20~30
    [23]韩力群.人工神经网络在控制中的应用及发展前景.北京轻工业学院学报,1996,14(1):32~36
    [24]Will M.J. Artifical neural networks in process engineering. IEE Proc-D, 1991, 138(3):256~266
    [25]Schlang Martin. Neural networks in real world applications. IEEE EXPERT, 1996, 11(4): 8~9
    [26]吴简彤,王建华.神经网络技术及其应用.哈尔滨:哈尔滨工程大学出版社,1998.50~60
    [27]杨奕若,王煦法.利用主元分析与神经网络进行人脸识别.电子技术应用,1998,6(3):21~22
    [28]张钹,张铃.人工神经网络的设计方法.清华大学学报,1998,38(S1):1~4
    [29]Hagan M T, Demouth H B, Beale M H. Neural Network Design. Boston, MA: PWS Publishing, 1996.20~35
    [30]王旭东,邵惠鹤.RBF神经网络理论及其在控制中的应用.信息与控制,1997,26(4):272~284
    [31]戚德虎,康继昌.BP神经网络的设计.计算机工程与设计,1998,19(2):48~50
    [32]李玉鉴.前馈神经网络中隐层神经元数目的一种直接估计方法.计算机学报,1999,22(11):1024~1028
    [33]魏海坤,宋文忠,徐嗣鑫.基于结构分解的神经网络设计算法.自动化学报,2001,27(2):276~279
    
    
    [34]张晓东,王伟,王小刚.选矿过程神经网络粒度软测量方法的研究.控制理论与应用,2000,19(1):85~88
    [35]韩璞,王东风,翟永杰.基于神经网络的火电厂烟气含氧量软测量.信息与控制,2001,30(2):189~192
    [36]Wang Yalin, Gui Weihua, Chen Xiaofang, et al. Neural network modeling for composition prediction of Pb-Zn sinter in imperial smelting process. Proceedings of the Annual Chinese Automation Conference in the UK. Derby, England, 1999(Sept): 21~24
    [37]G. Btoch, E Sirou, V. Eustache. Neural Intelligent Control for a Steel Plant. IEEE Trans. on Neural Networks, 1997, 8(4): 910~918
    [38]Y. Z. Lu, S. W. Marhaiard. Development and Application of an Integrated Neural System for an HDCL. IEEE Trans. on Neural Networks, 1997, 8(6): 1328~1337
    [39]Yarlagadda Prasad K D V. Development of an integrated neural network system for prediction of process parameters in metal injection moulding. Journal of Materials Processing Technology, 2002, 130:315~320
    [40]王雅琳,桂卫华,阳春华,等。自适应监督式分布神经网络及其工业应用.控制与决策,2001,16(5):549~552
    [41]杜玉晓,吴敏,桂卫华,等.铅锌烧结过程透气性的混沌遗传算法神经网络模型.见:第十三届中国过程控制会议论文集.广州:华南理工大学出版社,2002.205~209
    [42]Sungzoon. Reliable roll force prediction in cold mill using multiple neural networks. IEEE Trans. on Neural Networks, 1997, 21 (9): 981~996
    [43]吴敏,岑丽辉.铅锌烧结过程透气性预测与操作优化系统.见:第十三届中国过程控制会议论文集.广州:华南理工大学出版社,2002.108~112
    [44]M. Riedmiller, H. Braun. A direct adaptive method for faster backpropagation learning: the RPROP algorithm. In: E.H.Ruspini(Ed.), Proceeding of the IEEE International Conference on Neural Networks, New York: IEEE Press, 1993, 586~591
    [45]应苏如,桂卫华,李勇刚,等.智能集成模型在苛性比值软测量中的应用.见:中南工业大学学报第十四届中国过程控制学术年会暨第三届全国故障诊断与安全性学术会议论文专辑.长沙:中南工业大学出版社,2003.4~7

© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号

地址:北京市海淀区学院路29号 邮编:100083

电话:办公室:(+86 10)66554848;文献借阅、咨询服务、科技查新:66554700